Why I Let a Robot Manage My Money

Building an Automated Trading System

Prabuddha Tamhane

Land Acknowledgement

I wish to acknowledge that we gather and study on the traditional, ancestral, and unceded lands of the xʷməθkʷəy̓əm (Musqueam) and Sḵwx̱wú7mesh (Squamish) Nations.

I am thankful for the privilege to learn, work, and grow on this land.

The Problem: Humans Are Terrible Investors

In early 2024, I watched my stock drop 15% in one day.

I panicked. I sold everything.

Next week? It bounced back 20%.

Why do we do this? Because we’re human:

  • We panic when prices drop
  • We get greedy when prices rise
  • We chase trends and follow tips

Most individual investors underperform a simple index fund.

The Solution: Let a Robot Decide

What I Built

Paper Trader AI - Three strategies competing head-to-head:

  1. Momentum - Follow the trend
  2. XGBoost - ML pattern recognition
  3. LSTM - Deep learning predictions

All trading S&P 500 stocks with fake money, real prices.

Think of it as a flight simulator for investing.

Live Dashboard

Paper Trader Dashboard

Key Takeaways

  1. Automation removes emotion, but not risk
    • The bot won’t panic sell, but markets still change
  1. Simple strategies often beat complex ones
    • My basic momentum strategy outperformed fancy ML models
  1. Transaction costs are the silent killer
    • That “great” strategy might just be feeding the brokers

The Bottom Line

Now when my portfolio drops, I remind myself:

The robot wouldn’t care. Maybe I shouldn’t either.

Thank You!

Questions?

🔗 Live Dashboard: paper-trader-ai.streamlit.app

📂 GitHub: github.com/PAT0216/paper-trader